Hello there I'm Andrew Osborn, a Staff Software Engineer with 11+ years of experience building production AI systems where performance, reliability, and scalability aren't just goals they're requirements. I enjoy solving complex engineering problems from architecture to deployment, turning ambitious ideas into real-world products that millions of people can depend on. My expertise spans multimodal AI, large language models, distributed systems, and real-time inference across cloud, edge, and on-device environments. I specialize in designing low-latency AI pipelines, building intelligent agent workflows, optimizing model performance, and creating resilient backend systems that remain fast, secure, and reliable under demanding workloads. Whether it's fine-tuning inference, architecting scalable APIs, or developing evaluation frameworks that improve model quality and trustworthiness, I focus on building software that's both technically sound and practical to operate in production. Beyond writing code, I enjoy simplifying complex systems, mentoring engineers, and collaborating across teams to deliver thoughtful solutions. I'm passionate about building products that balance innovation with reliability, always keeping performance, user experience, and long-term maintainability at the center of every decision. If you're looking for someone who combines deep AI expertise with strong software engineering fundamentals and enjoys tackling challenging technical problems, I'd love the opportunity to connect and see what we can build together. Best regards, Andrew Osborn

Andrew M. Osborn

Hello there I'm Andrew Osborn, a Staff Software Engineer with 11+ years of experience building production AI systems where performance, reliability, and scalability aren't just goals they're requirements. I enjoy solving complex engineering problems from architecture to deployment, turning ambitious ideas into real-world products that millions of people can depend on. My expertise spans multimodal AI, large language models, distributed systems, and real-time inference across cloud, edge, and on-device environments. I specialize in designing low-latency AI pipelines, building intelligent agent workflows, optimizing model performance, and creating resilient backend systems that remain fast, secure, and reliable under demanding workloads. Whether it's fine-tuning inference, architecting scalable APIs, or developing evaluation frameworks that improve model quality and trustworthiness, I focus on building software that's both technically sound and practical to operate in production. Beyond writing code, I enjoy simplifying complex systems, mentoring engineers, and collaborating across teams to deliver thoughtful solutions. I'm passionate about building products that balance innovation with reliability, always keeping performance, user experience, and long-term maintainability at the center of every decision. If you're looking for someone who combines deep AI expertise with strong software engineering fundamentals and enjoys tackling challenging technical problems, I'd love the opportunity to connect and see what we can build together. Best regards, Andrew Osborn

Available to hire

Hello there
I’m Andrew Osborn, a Staff Software Engineer with 11+ years of experience building production AI systems where performance, reliability, and scalability aren’t just goals they’re requirements. I enjoy solving complex engineering problems from architecture to deployment, turning ambitious ideas into real-world products that millions of people can depend on.
My expertise spans multimodal AI, large language models, distributed systems, and real-time inference across cloud, edge, and on-device environments. I specialize in designing low-latency AI pipelines, building intelligent agent workflows, optimizing model performance, and creating resilient backend systems that remain fast, secure, and reliable under demanding workloads. Whether it’s fine-tuning inference, architecting scalable APIs, or developing evaluation frameworks that improve model quality and trustworthiness, I focus on building software that’s both technically sound and practical to operate in production.
Beyond writing code, I enjoy simplifying complex systems, mentoring engineers, and collaborating across teams to deliver thoughtful solutions. I’m passionate about building products that balance innovation with reliability, always keeping performance, user experience, and long-term maintainability at the center of every decision.
If you’re looking for someone who combines deep AI expertise with strong software engineering fundamentals and enjoys tackling challenging technical problems, I’d love the opportunity to connect and see what we can build together.
Best regards,
Andrew Osborn

See more

Language

Work Experience

Staff Software Engineer at Google
April 1, 2023 - Present
Architected the core multimodal runtime for Android XR from early platform foundation through Gemini integration, fusing live camera, microphone, and spatial context into model queries powering hands-free assistant features (visual Q&A, live translation, navigation) across headsets and glasses. Designed a hybrid on-device/cloud inference path that routes lightweight requests to on-device Gemini Nano on the Snapdragon NPU and escalates complex reasoning to cloud, targeting interactive response under ~300ms for common vision queries. Built an agentic action-routing layer that lets the assistant invoke Android app functions via explicit user confirmation rather than autonomous actions. Drove LLM evaluation/qualification across offline and online evals with grounding/hallucination checks, latency/cost monitoring, and regression dashboards across two device form factors ahead of launch. Added privacy boundaries for always-on sensors, including client-side filtering before frames reach the m
Senior Software Engineer at Google
November 1, 2020 - April 1, 2023
Led the real-time multimodal pipeline for AR translation glass prototypes, streaming on-glasses audio and video to ML services for speech recognition, translation, and OCR, with results rendered as in-lens captions. Engineered incremental ASR and streaming translation (chunked audio, partial hypotheses) to hold end-to-end caption latency under ~1 second for more natural live conversation. Built the edge/cloud split for battery-bound hardware (wake-word and UI local; heavy inference offloaded) with graceful degradation on dropped connectivity and an audio-only mode for longer outdoor sessions. Implemented privacy-by-design controls (recording indicator LED, capture restrictions, face/license-plate scrubbing) that carried the prototypes through controlled field trials. Set up translation-quality and latency dashboards from field telemetry to prioritize pipeline work; partnered with North/Focal team (post-acquisition), ML, and privacy reviewers.
Software Engineer III at Google
November 1, 2018 - November 1, 2020
Designed and owned the split-compute ML runtime for AR wearables (Glass Enterprise Edition 2), including dynamic inference routing that decides whether an on-glasses model runs on glasses, on a paired phone, or in cloud based on latency, battery, thermal headroom, and privacy (basis for issued patents). Built the on-device inference path on TensorFlow Lite with quantized/pruned models tuned to the Glass EE2 Snapdragon XR1 power/thermal budget with automatic fallback to local-only mode on disconnect. Implemented sensor sandboxing to prevent untrusted app code from spoofing sensor readings, separating trusted vs untrusted wearable peripherals. Defined model and sensor APIs and the Protocol Buffers/gRPC streaming contract between headset and companion device, integrated with Android framework ML runtimes. Compressed features on-device before offload to cut wireless data volume and keep within wearable bandwidth constraints; tuned runtime against real enterprise workloads in Glass EE2 pilo
Software Engineer II at Google
January 1, 2017 - November 1, 2018
Built device-side telemetry collectors and analysis pipeline for Daydream VR (frame timing, motion-to-photon latency, thermal, crash data), measuring against the 90fps/~20ms comfort budget VR demands. Implemented unsupervised anomaly detection (isolation-forest style models) over historical telemetry to flag latency spikes and frame drops and performance drift before release, with human-in-the-loop triage to hold down false positives. Developed regression dashboards comparing FPS, crash rate, and stability across app builds, OS releases, and devices (Daydream View, Mirage Solo/World Sense 6DoF). Wired quality checks into nightly CI/CD so each build ran on a device pool with telemetry auto-ingested, making quality reports part of the release checklist.
Software Engineering Intern at Google
May 1, 2016 - August 1, 2016
Built an analytics framework collecting Android system-health data across multiple Google apps, scaled to 50+ devices with automated collection and build-over-build regression visualizations.
Software Engineer Intern at Simply Hire
June 1, 2015 - August 1, 2015
Implemented geolocation workflow components for job postings using Google Maps and Google Places APIs.
Software Development Intern at West Health
June 1, 2014 - June 1, 2015
Built healthcare workflow and integration services using Java, JBoss, Hibernate, WSO2, JMS, XML/XSD, AWS S3/EC2, and automated test scripts.

Education

B.S. in Computer Engineering at Georgia Institute of Technology
January 1, 2012 - January 1, 2016

Qualifications

Add your qualifications or awards here.

Industry Experience

Software & Internet, Computers & Electronics, Telecommunications, Gaming

Hire a AI Prompt Engineer

We have the best ai prompt engineer experts on Twine. Hire a ai prompt engineer in Florida today.